1. PyMW - Master Worker Computing for Python

This is the documentation for PyMW (Python Master Worker). PyMW is a Python module for use in master-worker style parallel computation. PyMW provides a common API to multiple computation environments. This allows users to write a single Python program that can easily scale from a single computer to a worldwide computing grid.

The focus of PyMW is providing tools for simple, user-friendly parallel computing. Although high performance is a goal of PyMW, the main goal is simplicity.

1.1. Computations suitable for PyMW

Parameter Sweep

A parameter sweep is where the same computation is repeatedly performed using different inputs each time. This is often used to sweep through a range of parameters and find the relation between inputs and outputs.

Monte Carlo

The Monte Carlo method uses random numbers to account for uncertainty in a computation. The computation is run a large number of times and statistics are gathered. For example, the Monte Carlo method is used in physics simulations where particle decay is random.

Genetic Algorithms/Optimization Techniques

Genetic algorithms attempt to optimize a function with regards to a set of inputs. In this case, master-worker parallelization may work well in evaluating each possible solution.

1.2. Types of computations not suitable for PyMW

PyMW is not well suited for computations with frequent communication and sharing of data between workers.

Table Of Contents

Previous topic

Welcome to PyMW’s documentation!

Next topic

2. Installation

This Page